Bilingual Education Programs: The Main Models Compared

Telo AI helps school districts improve speaking outcomes for English Learners and support bilingual education programs through conversational AI and practical tools for educators.

A teacher with small groups of children, illustrating bilingual education programs

Estimated reading time: 7 minutes

Bilingual education programs come in several models, and they differ most in one thing: whether the goal is to move students into English quickly or to build lasting bilingualism. This guide compares the main program models side by side, explains how they relate to English-only options, and gives a simple framework for choosing, so district leaders and families can tell them apart.

Table of contents

What Are the Main Bilingual Education Programs?

Quick answer: the main models are transitional bilingual education, developmental (maintenance) bilingual education, and dual language (including two-way immersion). They differ in goal: transitional programs use the home language as a temporary bridge to English, while developmental and dual language programs aim for full bilingualism and biliteracy. Sheltered approaches like SIOP can be used within any of them.

The Models Compared

ModelGoalHome language
Transitional (early or late exit)Move to English-only instructionTemporary bridge, then phased out
Developmental / maintenanceBilingualism and biliteracy for one language groupDeveloped and maintained
Dual language (one-way)Full bilingualism for mostly one groupDeveloped alongside English
Two-way immersionBilingualism for native English and native partner-language speakers togetherBoth languages developed

The Underlying Divide: Additive vs Subtractive

Behind the model names is a single organizing idea. Subtractive programs (transitional) use the home language temporarily and replace it with English, so students usually end up monolingual. Additive programs (developmental, dual language, two-way immersion) build the home language alongside English toward full bilingualism. Knowing whether a model is additive or subtractive tells you, at a glance, what students will walk away with.

Where English-Based Models Fit

Not every English Learner is in a bilingual program. Where students speak many different home languages, districts use English-based models instead: ESL or ENL, Structured English Immersion (SEI), and newcomer programs for recently arrived students. These are not bilingual education (they do not teach content in the home language), but they often use the same sheltered techniques. See bilingual education vs ESL for the full contrast.

How Districts Choose

The choice depends on goals, demographics, and staffing:

  • Shared home language? A large, stable group speaking one language makes dual language or maintenance viable; very mixed languages point to ESL/ENL.
  • Program goal? Fast English transition favors transitional or SEI; lasting bilingualism favors dual language.
  • Staffing? Dual language requires certified bilingual teachers, the hardest role to fill.
  • Commitment? Additive models pay off over years, so they need continuity across grades.

See what is bilingual education for the bigger picture.

Early-Exit and Late-Exit: Two Kinds of Transitional Program

Transitional bilingual education is not a single design; it comes in two forms that differ sharply in how long they keep the home language. Early-exit programs move students into all-English instruction quickly, typically within two to three years, using the home language mainly to ease the initial transition and prevent students from falling behind in content while they learn English. Late-exit programs, sometimes called developmental transitional, keep home-language instruction longer, often through fifth or sixth grade, and give it a larger share of the day before students move fully into English.

The distinction matters because the two produce different results. Early-exit programs get students into the mainstream sooner but provide less time to develop academic language in the home language, which can limit the deeper literacy that transfers between languages. Late-exit programs invest more years in the home language, and longitudinal research has generally found that students who receive more sustained home-language instruction reach stronger long-term academic outcomes than those exited quickly. For a district, the choice between them is really a choice about how much it values speed of transition versus depth of bilingual academic foundation, and it should be made with the understanding that language and literacy built in one language support the other rather than competing with it.

90/10 and 50/50: How Dual Language Programs Split the Day

Dual language and two-way immersion programs are defined not only by their goal of full bilingualism but by how they divide instructional time between the two languages, and the two common patterns are worth understanding. In a 90/10 model, the program begins in the early grades with about 90 percent of instruction in the partner language, Spanish for example, and only 10 percent in English, then gradually shifts toward a balance by the upper elementary grades. The logic is that English is already dominant in the surrounding environment, so front-loading the partner language protects it during the years when a strong foundation is easiest to build. In a 50/50 model, instruction is split roughly evenly between the two languages from the start, either by subject, by time of day, or by alternating days.

Neither split is universally better; the right one depends on the community, the availability of materials and teachers in the partner language, and the program’s goals. What both share is the expectation that students will spend substantial time actually using each language, not just receiving instruction in it. This is where many programs quietly fall short, because dividing the schedule is easier than ensuring each student gets enough active speaking practice in both languages, which is the input that turns time-on-language into genuine bilingualism.

Frequently Asked Questions

What are the types of bilingual education programs?

Transitional (early or late exit), developmental/maintenance, and dual language, including two-way immersion.

How do they differ?

Mainly by goal: transitional moves students to English and phases out the home language; developmental and dual language build lasting bilingualism.

What is the difference between additive and subtractive programs?

Additive programs develop the home language alongside English toward bilingualism; subtractive programs phase the home language out, so students usually end up monolingual in English.

Which model is most effective?

For long-term academic outcomes and bilingualism, dual language generally performs best, though it requires more resources and consistency.

Where does SIOP fit?

SIOP is a sheltered-instruction model that can be used within any bilingual or English-based program to make content comprehensible.

What is the difference between early-exit and late-exit programs?

Early-exit transitional programs move students into all-English instruction within about two to three years, using the home language briefly to ease the transition. Late-exit programs keep home-language instruction longer, often through fifth or sixth grade, giving it a larger share of the day before full transition to English.

What do 90/10 and 50/50 mean in dual language programs?

They describe how instructional time is split between the two languages. A 90/10 model starts with about 90 percent in the partner language and 10 percent in English, shifting toward balance by upper elementary. A 50/50 model splits time roughly evenly from the start, by subject, time of day, or alternating days.

Which model produces the best long-term outcomes?

Longitudinal research generally finds that additive programs, especially dual language and late-exit models that sustain home-language instruction, produce the strongest long-term academic outcomes for English Learners, because literacy built in one language supports the other. These models require more resources and continuity across grades.

Does the model guarantee results?

No. Any model depends on students actively using the target languages. Dividing the schedule is easier than ensuring each student gets enough speaking practice in both languages, which is the input that turns time-on-language into genuine bilingualism.

Conclusion

Bilingual education programs range from transitional bridges to full dual language immersion, and the right choice follows the district’s goal for students’ languages. Whatever the model, its success rests on students getting enough meaningful practice, especially speaking, in the languages it aims to build.

Sources: U.S. Department of Education, NCELA; Center for Applied Linguistics.

The Constraint That Outlasts the Model You Pick

Choosing among transitional, developmental, and dual language models feels like the decisive move, and it matters. But a district can select the perfect model on paper and still underdeliver, because every one of these designs runs on the same fuel: students actively using the languages the program targets.

That is the constraint no model choice escapes. Whether the goal is a fast transition or full biliteracy, learners need speaking practice at a level a single teacher cannot supply to a whole class, and in additive models they need it in two languages at once.

Districts that get results treat practice capacity as part of the model decision, not an afterthought. Telo AI is one example, with separate English, Spanish, and French apps that provide practice in whichever language or languages the chosen model targets.

See how districts back any model with enough practice: https://mytelo.ai/how-telo-works/

With that capacity in place, the model on paper becomes the program students actually experience. The right model is necessary, but it is the practice behind it that decides whether the program works.

See how Telo works in the classroom

Learn how Telo helps English Learners practice speaking at their own level while giving teachers real-time insights.